• Journal of Korean Association for Spatial Structures
• /
• v.21 no.2
• /
• pp.99-110
• /
• 2021

This study presents the estimation of crack depth by analyzing temperatures extracted from thermal images and environmental parameters such as air temperature, air humidity, illumination. The statistics of all acquired features and the correlation coefficient among thermal images and environmental parameters are presented. The concrete crack depths were predicted by four different machine learning models: Multi-Layer Perceptron (MLP), Random Forest (RF), Gradient Boosting (GB), and AdaBoost (AB). The machine learning algorithms are validated by the coefficient of determination, accuracy, and Mean Absolute Percentage Error (MAPE). The AB model had a great performance among the four models due to the non-linearity of features and weak learner aggregation with weights on misclassified data. The maximum depth 11 of the base estimator in the AB model is efficient with high performance with 97.6% of accuracy and 0.07% of MAPE. Feature importances, permutation importance, and partial dependence are analyzed in the AB model. The results show that the marginal effect of air humidity, crack depth, and crack temperature in order is higher than that of the others.

• Journal of Biomedical Engineering Research
• /
• v.43 no.5
• /
• pp.353-360
• /
• 2022

Laryngeal disease harms quality of life, and laryngoscopy is critical in identifying causative lesions. This study extracts and analyzes using radiomics quantitative features from the lesion in laryngoscopy images and will fit and validate a classifier for finding meaningful features. Searching the region of interest for lesions not classified by the YOLOv5 model, features are extracted with radionics. Selected the extracted features are through a combination of three feature selectors, and three estimator models. Through the selected features, trained and verified two classification models, Random Forest and Gradient Boosting, and found meaningful features. The combination of SFS, LASSO, and RF shows the highest performance with an accuracy of 0.90 and AUROC 0.96. Model using features to select by SFM, or RIDGE was low lower performance than other things. Classification of larynx lesions through radiomics looks effective. But it should use various feature selection methods and minimize data loss as losing color data.

• Journal of Korea Society of Industrial Information Systems
• /
• v.14 no.3
• /
• pp.1-14
• /
• 2009

In this paper, we propose a feature vector and its applying method which can be utilized for the extraction of the car license plate region. The proposed feature vector is extracted from direction code histogram of edge direction of gradient vector of image. The feature vector extracted is forwarded to the MLP classifier which identifies character and garbage and then the recognition of the numeral and the location of the license plate region are performed. The experimental results show that the proposed methods are properly applied to the identification of character and garbage, the rough location of license plate, and the recognition of numeral in license plate region.

• Journal of Forest and Environmental Science
• /
• v.40 no.1
• /
• pp.15-23
• /
• 2024

This research aimed to assess the possibility of detecting forest degradation using time-series satellite imagery and three different deep learning-based change detection techniques. The dataset used for the deep learning models was composed of two sets, one based on surface reflectance (SR) spectral information from satellite imagery, combined with Texture Information (GLCM; Gray-Level Co-occurrence Matrix) and terrain information. The deep learning models employed for land cover change detection included image differencing using the Unet semantic segmentation model, multi-encoder Unet model, and multi-encoder Unet++ model. The study found that there was no significant difference in accuracy between the deep learning models for forest degradation detection. Both training and validation accuracies were approx-imately 89% and 92%, respectively. Among the three deep learning models, the multi-encoder Unet model showed the most efficient analysis time and comparable accuracy. Moreover, models that incorporated both texture and gradient information in addition to spectral information were found to have a higher classification accuracy compared to models that used only spectral information. Overall, the accuracy of forest degradation extraction was outstanding, achieving 98%.

• Transactions of Materials Processing
• /
• v.33 no.4
• /
• pp.248-254
• /
• 2024

The triply periodic minimal surface (TPMS) structure is characterized by a high surface-to-volume (S/V) ratio and the separated internal structure for flow. Combining the different TPMS structures can provide unique flow and strength characteristics. This paper investigates the effects of dimension, density and arrangement of the unit cell of the TPMS on contact and flow areas of combined TPMS structures. Several representative TPMS structures, including primitive, gyroid and diamond structures, are adopted to design gradient and heterogeneous types TPMS structures. The estimation method of contact and flow areas using an image processing technique is proposed. Python software is used to predict contact and flow area. The influence of the combination method of TPMS on contact and flow areas in the contact surface of combined TPMS structures with different shapes is investigated. Based on the results of the investigation, an appropriate combination method of TPMS structures is discussed.

• Journal of Korea Water Resources Association
• /
• v.47 no.6
• /
• pp.537-546
• /
• 2014

Surface image velocimetry was introduced as an efficient and sage alternative to conventional river flow measurement methods during floods. The conventional surface image velocimetry uses a pair of images to estimate velocity fields using cross-correlation analysis. This method is appropriate to analyzing images taken with a short time interval. It, however, has some drawbacks; it takes a while to analyze images for the verage velocity of long time intervals and is prone to include errors or uncertainties due to flow characteristics and/or image taking conditions. Methods using spatio-temporal images, called STIV, were developed to overcome the drawbacks of conventional surface image velocimetry. The grayscale-gradient tensor method, one of various STIVs, has shown to be effectively reducing the analysis time and is fairly insusceptible to any measurement noise. It, unfortunately, can only be applied to the main flow direction. This means that it can not measure any two-dimensional flow field, e.g. flow in the vicinity of river structures and flow around river bends. The present study aimed to develop a new method of analyzing spatio-temporal images in two-dimension using cross-correlation analysis. Unlike the conventional STIV, the developed method can be used to measure two-dimensional flow substantially. The method also has very high spatial resolution and reduces the analysis time. A verification test using artificial images with lid-driven cavity flow showed that the maximum error of the method is less than 10 % and the average error is less than 5 %. This means that the developed scheme seems to be fairly accurate, even for two-dimensional flow.

• Journal of the Korean Society of Radiology
• /
• v.10 no.2
• /
• pp.73-79
• /
• 2016

The purpose of this analysis is to compare 2D T1 FEE and 3D T1 THRIVE for demonstration of the pancreas. A total of 85(45 men, 40 women; 58 years) PACS network datum were analysis clinically indicated pancreas MRI at 1.5 T. The SNRs and CNRs of 3D T1 THRIVE(SNR: $46.42{\pm}0.67$, CNR: $28.16{\pm}0.50$) showed significantly higher values than those from 2D T1 FEE(SNR: $53.84{\pm}1.20$, CNR: $35.48{\pm}0.70$), p<0.05, The image quality of the 3D T1 THRIVE($2.63 {\pm}0.14$) was significantly superior to that with the 2D T1 FEE($2.2{\pm}0.05$), but 3D T1 THRIVE revealed several artifacts resulting in poor quality. In conclusion, The 3D T1 THRIVE technique with a 1.5 T resulting in improved SNRs, CNRs and image quality was demonstrated.

• Journal of the Korea Society of Computer and Information
• /
• v.19 no.10
• /
• pp.63-70
• /
• 2014

Process of recognizing objects in binary images consists of image segmentation and pattern matching. If binary objects in the image are assumed to be separated, global features such as area, length of perimeter, or the ratio of the two can be used to recognize the objects in the image. However, if such an assumption is not valid, the global features can not be used but local features such as points or line segments should be used to recognize the objects. In this paper points with large curvature along the perimeter are chosen to be the feature points, and pairs of points selected from them are used as local features. Similarity of two local features are defined using elastic deformation energy for making the lengths and angles between gradient vectors at the end points same. Neighbour support value is defined and used for robust recognition of partially occluded binary objects. An experiment on Kimia-25 data showed that the proposed algorithm runs 4.5 times faster than the maximum clique algorithm with same recognition rate.

• Investigative Magnetic Resonance Imaging
• /
• v.21 no.2
• /
• pp.71-81
• /
• 2017

Purpose: To compare three, motion-resistant, T1-weighted MR sequences on the hepatobiliary phase for gadoxetic acid-enhanced MR imaging of the liver. Materials and Methods: In this retrospective study, 79 patients underwent gadoxetic acid-enhanced, 3T liver MR imaging. Fifty-nine were examined using a standard protocol, and 20 were examined using a motion-resistant protocol. During the hepatocyte-specific phase, three MR sequences were acquired: 1) gradient recalled echo (GRE) with controlled aliasing in parallel imaging results in higher acceleration (CAIPIRINHA); 2) radial GRE with the interleaved angle-bisection scheme (ILAB); and 3) radial GRE with golden-angle scheme (GA). Two readers independently assessed images with motion artifacts, streaking artifacts, liver-edge sharpness, hepatic vessel clarity, lesion conspicuity, and overall image quality, using a 5-point scale. The images were assessed by measurement of liver signal-to-noise ratio (SNR), and tumor-to-liver contrast-to-noise ratio (CNR). The results were compared, using repeated post-hoc, paired t-tests with Bonferroni correction and the Wilcoxon signed rank test with Bonferroni correction. Results: In the qualitative analysis of cooperative patients, the results for CAIPIRINHA had significantly higher ratings for streak artifacts, liver-edge sharpness, hepatic vessel clarity, and overall image quality as compared to, radial GRE, (P < 0.016). In the imaging of uncooperative patients, higher scores were recorded for ILAB and GA with respect to all of the qualitative assessments, except for streak artifact, compared with CAIPIRINHA (P < 0.016). However, no significant differences were found between ILAB and GA. For quantitative analysis in uncooperative patients, the mean liver SNR and lesion-to-liver CNR with radial GRE were significantly higher than those of CAIPIRINHA (P < 0.016). Conclusion: In uncooperative patients, the use of the radial GRE sequence can improve the image quality compared to GRE imaging with CAIPIRINHA, despite the data acquisition methods used. The GRE imaging with CAIPIRINHA is applicable for patients without breath-holding difficulties.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.4 no.3
• /
• pp.126-133
• /
• 2003

This paper is related to a morphological segmentation method for extracting the moving object in video sequence using global motion compensation and two-dimensional spatio-temporal entropic thresholding. First, global motion compensation is performed with camera panning vector estimated in the hierarchical pyramid structure constructed by wavelet transform. Secondly, the regions with high possibility to include the moving object between two consecutive frames are extracted block by block from the global motion compensated image using two-dimensional spatio-temporal entropic thresholding. Afterwards, the LUT classifying each block into one among changed block, uncertain block, stationary block according to the results classified by two-dimensional spatio-temporal entropic thresholding is made out. Next, by adaptively selecting the initial search layer and the search range referring to the LUT, the proposed HBMA can effectively carry out fast motion estimation and extract object-included region in the hierarchical pyramid structure. Finally, after we define the thresholded gradient image in the object-included region, and apply the morphological segmentation method to the object-included region pixel by pixel and extract the moving object included in video sequence. As shown in the results of computer simulation, the proposed method provides relatively good segmentation results for moving object and specially comes up with reasonable segmentation results in the edge areas with lower contrast.